35 f/2 Focus Recompose (rotate) focus plane shift

[dwbell]

Advertisement (gone after registration)

I thought I'd share a little diagram I did to answer I question I myself had.

 

M9, Summicron 35mm ASPH, assuming a "flat" plane of focus (which it's said to have).

 

BLUE;

Depth of field (3 parallel lines) focus on "eyes"

Near limit 67.8 cm

Far limit 72.4 cm

Total 4.56 cm

Angled lines = 54deg field of view (horizontal)

 

GREEN;

Having focused, rotate camera to place subject into vertical Rule of Thirds line (9.2deg)

Parallel lines same relationship as blue.

 

PINK;

Approximation of human head (mine!)

 

My summary is that if I focus and rotate-recompose at the minimum focus distance at the maximum aperture the eye's will still be in focus. This holds true until you basically have the head "inside" the outer third of the image (or "outside" the vertical thirds line). At that point you need to stop down or move backwards.

 

Drawn on CAD 1:1

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[jplomley]

Just shift left or right to keep the back parallel to your subject when composing for rule of thirds after focussing in the RF patch and you should be fine.

[dwbell]

Jeff,

 

Yes, this was why I did the exercise. I don't like to "pan" once I've got my framing as it changes my perspective. With testing I noticed that I could hold focus with the rotating method and was interested to discover what, mathematically, the "error" would be. My findings indicate that the error is not enough to warrant worrying about, assuming you don't go right to the edge of frame.

 

The numbers indicate, further, that if I put my subject right on the edge of the frame I need to move my head backwards 3cm to re-attain focus. In practice this is easily done.

 

Compose for visual effect, move patch over subject to focus (rotate), rotate back to required framing and if required (subject on the very edge of the frame) just move your head/camera back 2-3cm - about the length of my nose. Not much.

 

I found it useful to know what order of magnitude these movements were with this lens.

 

Kind regards,

DWBell

[yanidel]

That is an interesting way to look at it.

Other conclusion you coud draw from it is to set focus on the nose before you recompose so it remains in the middle of depth of field.

[dwbell]

Indeed. Worth noting also that if you had the right hand edge of the face on the right hand side of the frame (having focused centrally, always in my examples) you would need to open up to f/5.6 - f/8 to attain sufficient DOF to still hold the eye in focus. Not something most people bought good Leica glass to do! =)

[Guest mc_k]

You can't take the theoretical depth of field literally; just worry about the error.

 

Sure, you are talking about a 1/2 cm error at 0.7m. But if you redid the example for 2m, e..g. reframing from a head to a waist, the problem is more real. It's not hard to get a simple formula for the error in terms of the distance and focal length (for rule-of-thirds, or any other placement).

[giordano]

Advertisement (gone after registration)

The "Barnack" depth of field calculator automatically calculates and displays a "focus recomposition circle" for this purpose.Here's what it shows for a 35mm lens at f/2 at 70cm, first with a traditional 0.3mm CoC and then with a hypercritical 2-pixel CoC:

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[dwbell]

mckeough_k - I'm afraid I don't understand what you said? I don't take the theory too literally and I was interested (not worried so much) about the error. Hence the diagrammatic approach. At 2m I've got about 40cm of DOF at f/2.0 with 35mm (yes, theoretical.) Does this not prove out in practice?

 

giordano - I think I'm tired this evening. What does the circle show me? That I can move (rotate) the patch 342mm away from the centre focus and the subject will still be in acceptable focus? I'll check if mine correlates but it looks about right?

[giordano]

What does the circle show me?

 

When you focus exactly on an object and then recompose, if the object remains within the circle it will also be within the depth of field. The size of the circle varies with the aperture, focal length, focus distance and circle of confusion.

[Guest mc_k]

mckeough_k - I'm afraid I don't understand what you said? I don't take the theory too literally and I was interested (not worried so much) about the error. Hence the diagrammatic approach. At 2m I've got about 40cm of DOF at f/2.0 with 35mm (yes, theoretical.) Does this not prove out in practice?

...

 

well you can try it out. Try measuring the depth of what's in focus a couple of different ways, and afterwards use the dof calculator. Is the measured depth independent of what you used to get it? Is the measured depth close to the predicted?

 

I think you are mixing something real--the error--with something not so real. If you're interested in the problem practically, you can get a mannequin head to shoot. If you're interested for its own sake, come up with something that will cover all cases, not just this one.

[dwbell]

Thank you for the clarification.

 

With respect though what you think I'm mixing or not is just your opinion, which of course is valid and you are entitled to. I was interested in how far to the edge of the frame at 35mm f/2 and 70cm I could push a persons head with focus rotate-recompose whilst maintaining acceptable focus on the eyes. Following empirical observations (it seemed to be a relative non problem at "normal" subject placement) I wanted a diagrammatic model to show me the order of magnitude / boundaries of the issue. I thought that this might be of interest to others, which it seems t have been.

 

I don't as yet need to come up with something for all cases, as I've happily learned there is a dof program that gives me the "focus recomposition circle," a much more elegant solution than my own. I've learnt something and got some great advice - a good percentage of the reason I've recently entered into RF photography. I'm a happy man.

 

I appreciate that there's a difference between theory and practice. If I understand you correctly then in that point I wholeheartedly agree with you.

[Guest mc_k]

... I've happily learned there is a dof program that gives me the "focus recomposition circle," a much more elegant solution than my own...

 

it's the same diagram, but you are just rotating in a plane.

 

And now you have to determine which of the two circles, if either, is correct.

 

Just calculate the error for some cases you might come across. If it's really small ignore it; if it's big, compensate it. Why zone focus when you can focus?

[giordano]

it's the same diagram, but you are just rotating in a plane.

 

And now you have to determine which of the two circles, if either, is correct.

 

What do you mean by "which of the two circles"? Barnack only shows you one: I posted two pictures to show that its size depends on the circle of confusion you choose.

 

Remember that the depth of field markings on the lens are based on assumptions about film and lens quality and size of enlargements that date from the 1930s (a 0.03mm circle of confusion). If you want things to look really sharp in a big print from an M9 you need to calculate depth of field using a much smaller circle of confusion (there have been many threads about this). The smaller circle in my post is based on a 2-pixel circle of confusion (on an M9), i.e. the maximum detail the sensor can provide. In practice, one would choose a circle of confusion somewhere between the two.

[dwbell]

Just calculate the error for some cases you might come across. If it's really small ignore it; if it's big, compensate it. Why zone focus when you can focus?

 

Yes, good advice. Something I intend to do as and when I come across the issue.

 

giordano - yep, all understood. Thanks again for the link and explanation.

[Guest mc_k]

What do you mean by "which of the two circles"? Barnack only shows you one: I posted two pictures to show that its size depends on the circle of confusion you choose.

...

 

that's what I meant; the circles/zones will always be a little dubious because they are based on a guess for that value.

[Guest mc_k]

I had another look at the diagram, which I know took some work--thanks.

...

Angled lines = 54deg field of view (horizontal)

...

o.k., from published

 

Indeed. Worth noting also that if you had the right hand edge of the face on the right hand side of the frame (having focused centrally, always in my examples) you would need to open up to f/5.6 - f/8 to attain sufficient DOF to still hold the eye in focus...

 

o.k.

 

...

Having focused, rotate camera to place subject into vertical Rule of Thirds line (9.2deg)...

 

how did you get 9.2 deg?

 

 

...

The numbers indicate, further, that if I put my subject right on the edge of the frame I need to move my head backwards 3cm to re-attain focus...

 

how did you get 3cm?

[Guest mc_k]

previous installment of this thread ... http://www.l-camera-forum.com/leica-forum/leica-m9-forum/119220-wa-focussing-advice-please.html

[mjh]

that's what I meant; the circles/zones will always be a little dubious because they are based on a guess for that value.

Everything one can say about depth of field ultimately depends on the chosen CoC; nothing new here. If the ultimate arbitrariness of the CoC limit troubles you, you would have to give up on the concept of depth of field altogether. Or you would have to spend some time finding a CoC value giving acceptable results (acceptable to you that is).

[Guest mc_k]

Everything one can say about depth of field ultimately depends on the chosen CoC; nothing new here. If the ultimate arbitrariness of the CoC limit troubles you, you would have to give up on the concept of depth of field altogether. Or you would have to spend some time finding a CoC value giving acceptable results (acceptable to you that is).

 

o.k., it is easy enough to shoot a ruler and calculate a value for the uncertainty yourself.

 

I do think the original poster proved his point, that reframing matters some times and not others... how would you prove the point, in the same way?

[Advertisement]

Hallo,

Look here: Leica M

[Guest mc_k]

...If the ultimate arbitrariness of the CoC limit troubles you, you would have to give up on the concept of depth of field altogether...

 

btw, you don't need the value of a constant in a physical formula to make use of the formula